Is IoT Somehow Psychic?

Sensing What?

Lone Star has years of analytics experience. We are often amused by magical thinking and confusion about IoT sensors. There are at least three kinds of this confusion across all types of IoT. It’s worth thinking about each of them.  Together they amount to asking the question, “Sensing what?”

The first type of confusion is failure to ask, “What does this sensor do?”

This might seem silly. IT SENSES, DUMMY! But it would be wonderful if we knew how it generated measurements. For example, all (100%) of thermal sensors have time lag.

Precise thermal sensors may be S.L.O.W. If fast, they may have precision problems, or be expensive. And cost matters. Most IoT applications can’t afford many sensors. We might use a fast fire protection sensor’s temperature data for other IoT purposes. But we must remember it may not be very accurate.

There is no magic temperature sensor with perfect accuracy and zero delay. For some IoT applications this may not matter. But we need to ask, “What does the sensor do?”

The second type of confusion is what a sensor is reporting.

We see this in vibration sensors.

Recently, we did a proof of concept. We were given a vibration data stream. It had interesting patterns. But more interesting was the response when we asked what the sensor was reporting. It took a while to get the answer, because there are many different types of “vibration sensors.”

• Displacement; a sensor measures distance – how far is a sensor moving back and forth?
• Velocity; a sensor measures speed – how fast is a sensor moving back and forth?
• Acceleration; a sensor measures G’s – usually to estimate forces in back and forth motion.

In a perfect world, with perfect sensors, and perfect data bandwidth, these are interchangeable. We know the equations to move from force to acceleration to velocity to position, and back.

But real sensors are limited. It may be impossible to use IoT position to estimate acceleration or force.

A third type of confusion is what a sensor’s data means.

In the proof of concept, the sensor was reporting “Gs.” We asked, ‘what kind of G’s?’ There are:

• Gs “RMS” – all acceleration regardless of direction, smoothed by “root mean squared” and always a positive number.
• Gs to snapshot acceleration in a time window (50% of the data should be positive numbers)
• Gs in a specific frequency band (looking for specific energy inputs) rather than broad band

There are many other ways to sense vibration as acceleration. And, the other vibration sensors (position, velocity) are diverse, too.

So, are your sensors psychic? Do they know what you meant to measure? Can they guess your hopes and dreams for IoT?

Sadly, no. Your sensor does not have ESP.

The AnaltyicsOS application design environment (AOS Architect) has a critical feature; sensor simulators. AOS Architect simulates different sensors and their limitations. It helps you understand how sensors and an IoT application might cope with real-world conditions, bandwidth limits, and other nasty realities.

So, for IoT analytics success we suggest three things:

1. Ask the three questions about your IoT sensor.
2. Don’t wait for ESP sensors. Start creating value with IoT now.
3. Work with applications developed to thrive in the real world; AnalyticsOS.

Or, you could pretend your sensors have ESP.